↓ Skip to main content
Dan Treccagnoli’s Portfolio

My First Foray into Agentic AI

Dan Treccagnoli
Author
Dan Treccagnoli

My First Foray into Agentic AI
#

I wanted to experiment with Agentic AI as a coding assistant to help me learn GDScript, the native programming language of the Godot Game Engine.

I did some research into achieving this using Gemini, and set to work writing a custom automation script in Python. The goal was to programmatically call the Gemini API to help me dynamically generate GDScript code and allow me to test the scripts directly within the Godot engine.

Overcoming Technical Hurdles
#

To make this workflow viable, I needed the assistant to maintain persistent context across interactions, allowing me to iterate on and refine a specific script within the same prompt chain. I relied heavily on API documentation and tutorials to nail down the correct syntax, iterating on my Python framework several times before getting successful, actionable results.

Key Takeaways & Skill Growth
#

  • Practical API Experience: I gained valuable, hands-on experience structuring API payloads, managing session state, and handling JSON responses.
  • Python Development: Building this tool allowed me to significantly flex and develop my burgeoning Python capabilities.
  • Interactive Workflows: Bridging an AI ecosystem with a game development environment opened up exciting new development possibilities.

This has honestly been the most interesting and fun experience I have had with AI so far!

Current Status & Future Outlook
#

My current iteration is operating on the free tier, which naturally introduces some capacity limits on API requests. Despite those boundary lines, I have already successfully produced several working GDScript modules. Moving forward, I plan to continue expanding this custom tool and experimenting with other workflows where AI can complement both my technical hobbies and professional projects.